Module:gu-IPA/sandbox

The following documentation is located at Module:gu-IPA/sandbox/documentation. ^[edit] Categories were auto-generated by Module:module categorization. ^[edit]

Useful links: root page • root page’s subpages • links • transclusions • testcases • sandbox of (diff)

Gujarati IPA pronunciation module. See {{gu-IPA}}.

Testcases

Module:gu-IPA/sandbox/testcases:

(refresh)

test_all:

	Text	Expected	Actual	Comments
	વખત (vakhat)	ˈʋə.kʰət̪	ˈʋə.kʰət̪
	દિવસ (divas)	ˈd̪ɪ.ʋəs	ˈd̪ɪ.ʋəs
	હદ (had)	ˈɦəd̪	ˈɦəd̪
	રહેવું (rahevũ)	ˈre̤.ʋũ	ˈɾə.ɦeː.ʋũ
	તડકો (taḍko)	ˈt̪əɖ.ko	ˈt̪əɖ.ko
	ભાષા (bhāṣā)	ˈbʱɑ.ʃɑ	ˈbʱɑ.ʃɑ
	છોકરું (chokrũ)	ˈt͡ʃʰok.rũ	ˈt͡ʃʰok.ɾũ
	ઘોડું (ghoḍũ)	ˈɡʱo.ɖũ	ˈɡʱo.ɖũ
	દૃષ્ટિ (dŕṣṭi)	ˈd̪rəʂʈi	ˈd̪ɾʊʂʈɪ
	દૃષ્ટિ (dŕṣṭi)	ˈd̪rʊʂʈi	ˈd̪ɾʊʂʈɪ
	ગુજરાતી (gujrātī)	ˈgʊ.d͡ʒ(ə).ˌɾɑ.t̪i	ˈɡud͡ʒ.ɾɑ.t̪i
	પ્રાણી (prāṇī)	ˈpɾɑ.ɳi	ˈpɾɑ.ɳi
	સાક્ષાત (sākṣāt)	ˈsɑ.kʃɑt̪	ˈsɑk.ʃɑt̪
	જમવાનું (jamvānũ)	ˈd͡ʒəm.ˌʋɑ.nũ	ˈd͡ʒəm.ʋɑ.nũ
	વિધર્મ (vidharma)	ʋɪ.d̪ʱəɾ.mə	ˈʋɪ.d̪ʱəɾ.mə
	અંગરખો (aṅgarkho)	ˈəŋ.gəɾ.kʱo	ˈəŋ.ɡəɾ.kʰo
	આંતરરાષ્ટ્રીય (āntarrāṣṭrīya)	ˈɑn̪.t̪əɾ.ɾɑʂ.ʈɾi.jə	ˈɑn.t̪əɾ.ɾɑʂ.ʈɾi.jə
	આંકડાશાસ્ત્ર (ā̃kaḍāśāstra)	ˈɑ̃.k(ə).ɖɑ.ʃɑs.t̪ɾə	ˈɑ̃.kə.ɖɑ.ʃɑs.t̪ɾə
	અહીંતહીં (ahī̃thī̃)	ˈə̤ĩ.ˌt̪ə̤ĩ	ˈə.ɦĩ.t̪ʰĩ
	ઐતિહાસિક (aitihāsik)	əɪ̭.t̪ɪ.ɦɑ.sɪk	ˈəɪ.t̪ɪ.ɦɑ.sɪk
	ઇસ્પિતાલ (ispitāl)	ˈɪs.pɪ.ˌt̪ɑl	ˈɪs.pɪ.t̪ɑl
	ઉષ્ણાગ્ર (uṣṇāgra)	ˈʊʃ.ɳɑ.gɾə	ˈuʃ.ɳɑɡ.ɾə
	ઓળખકર્તા (oḷkhakartā)	ˈo.ɭəkʱ.ˌkəɾ.t̪ɑ	ˈo.ɭkʰə.kəɾ.t̪ɑ

---

local export = {}

local lang = require("Module:languages").getByCode("gu")
local sc = require("Module:scripts").getByCode("Gujr")
local m_IPA = require("Module:IPA")

local gsub = mw.ustring.gsub
local gmatch = mw.ustring.gmatch
local find = mw.ustring.find

local correspondences = {
    ["ṅ"] = "ŋ", ["g"] = "ɡ", 
    ["c"] = "t͡ʃ", ["j"] = "d͡ʒ", ["ñ"] = "ɲ",
    ["ṭ"] = "ʈ", ["ḍ"] = "ɖ", ["ṇ"] = "ɳ",
    ["t"] = "t̪", ["d"] = "d̪",
    ["y"] = "j", ["r"] = "ɾ", ["v"] = "ʋ", ["l"] = "l̪",
    ["ś"] = "ʃ", ["ṣ"] = "ʂ", ["h"] = "ɦ",
    ["ḷ"] = "ɭ", ["f"] = "f", ["ġ"] = "ɣ", ["ḏ"] = "ð", ["ḇ"] = "β", 
    ["ṛ"] = "ɽ",

    ["a"] = "ə", ["ā"] = "ɑ", ["i"] = "ɪ",
    ["ī"] = "i", ["o"] = "o", ["e"] = "eː", ["ŕ"] = "ɾʊ",
    ["u"] = "u", ["ū"] = "u", ["ǒ"] = "ɔ", ["ɔ"] = "ɔ", ["ě"] = "ɛ", ["â"] = "æ",
    ["ä"] = "ə̤", ["ǟ"] = "a̤", ["ï"] = "i̤", ["ü"] = "ṳ",  ["ë"] = "ɛ̤", ["ö"] = "ɔ̤", 
    ["ॐ"] = "om", ["ḥ"] = "ʰ",
    [" "] = "‿ˈ", -- get rid of spaces
}

local surati= {
	["s"] = "ɦ",
}

local vowels = "aāiīuūoɔɛeæãā̃ẽĩī̃õũū̃ː"
local weak_h = "([gjdḍbṛnmaãāā̃eẽiĩīī̃uũūū̃oõː])h"
local aspirate = "([kctṭp])"
local syllabify_pattern = "([" .. vowels .. "])([^" .. vowels .. "%.]+)([" .. vowels .. "])"

local function find_consonants(text)
	local current = ""
	local cons = {}
	for cc in mw.ustring.gcodepoint(text .. " ") do
		local ch = mw.ustring.char(cc)
		if find(current .. ch, "^[kgṅcjñṭḍṇtdnpbmyrlvśṣsh]$") or find(current .. ch, "^[kgcjṭḍṇtdpb]h$") then
			current = current .. ch
		else
			table.insert(cons, current)
			current = ch
		end
	end
	return cons
end

local function syllabify(text)
	for count = 1, 2 do
		text = gsub(text, syllabify_pattern, function(a, b, c)
			b_set = find_consonants(b)
			table.insert(b_set, #b_set > 1 and 2 or 1, ".")
			return a .. table.concat(b_set) .. c end)
	end
	return text
end

local identical = "knlsfzθ"
for character in gmatch(identical, ".") do
	correspondences[character] = character
end

local function transliterate(text)
	return (lang:transliterate(text))
end

function export.link(term)
	return require("Module:links").full_link{ term = term, lang = lang, sc = sc }
end

function export.toIPA(text)
	local result = {}
	local translit = text
	if lang:findBestScript(text):isTransliterated() then
		translit = transliterate(text)
	end
	if not translit then
		error('The term "' .. text .. '" could not be transliterated.')
	end
	
	local translit = gsub(translit, "͠", "̃")
	local translit = gsub(translit, "%-", ".")
	local translit = gsub(translit, "ṣ([^ṭḍ])", "ś%1")
	
    local translit = syllabify(translit)
    local translit = gsub(translit, 'jñ', 'gy')
    local translit = gsub(translit, aspirate .. "h", '%1ʰ')
    local translit = gsub(translit, weak_h, '%1ʱ')
	local translit = gsub(translit, "%.ː", "ː.")
	
	for character in gmatch(translit, ".") do
		table.insert(result, correspondences[character] or character)
	end
	
	result = table.concat(result)
	local result = gsub(result, "ː̃", "̃ː")

	result = gsub(result, "%.‿", "‿")
    
	return "ˈ" .. result
end

function export.make(frame)
	local args = frame:getParent().args
	local pagetitle = mw.title.getCurrentTitle().text
	
	local p, results = {}, {}
	
	if args[1] then
		for index, item in ipairs(args) do
			table.insert(p, (item ~= "") and item or nil)
		end
	else
		p = { pagetitle }
	end
	
	for _, Gujarati in ipairs(p) do
		table.insert(results, export.toIPA(Gujarati))
	end
	
	return m_IPA.format_IPA_full { lang = lang, items = {{ pron = "/" .. table.concat(results, "/, /") .. "/" }}, }
end

return export